In today's medicine, many elderly patients have diminished kidney or liver function and take numerous medications. Adverse drug interactions and toxicities are common. SmartChart(R) is a program which allows nurses and physicians to efficiently enter real time clinical data into PC's from the bedside. For the first the paper chart is generated through a microcomputer rather than by pen or dictation. During Phase I we will first modify SmartChart so that therapy entries of all patients refer to a central list which will automatically expand as new drugs are used and entered. We will then develop a drug monitoring program (SmartDose) to use the complete, accurate and timely clinical information to estimate proper drug dosages and monitor for potential drug interactions. Each time a therapy is changed or entered, a program will calculate the proper dosage according to a number of formulas, determine whether there are interactions with other drugs, and display messages if the drug prescription could cause clinical problems. The database will be created by simplifying the Stanford drug interaction database and adding generally accepted drug dosing and kinetic formulas for each drug title. Accuracy and utility of messages will be tested using clinical information from a dialysis patient population. In Phase II, we will test neural network simulation programs to see whether physicians and nurses could eventually "train" a computer to monitor drug use in their own specialty units and hospitals. We will also test Bayesian methods of drug dose calculation and prediction.